Map Clustering Is Not My Favorite

Point datasets have two distinct modes of visualization. First is an aggregate view which serves to show you the trends and spatial distribution. Second is the individual view showing details about the point itself, its attributes, etc.

Clustering (for all its faults) is the only off-the-shelf technique for seamlessly switching between these two modalities without having to change the underlying data representation. Need more detail? Zoom in. And the zoom level is adaptive so it works with any scale.

There are better aggregation techniques (summing to a hexagonal grid, heatmaps, etc) but they generally require a separate calculation (possibly server side) and then switching to the raw source for the individual point view, either manually or at some hardcoded zoom level. It's not the same experience - it feels like two separate map layers instead of one integrated clustered layer.

This is mostly a matter of what's available in the mapping libraries. You could imagine building an alternative to clustering that calculates a heatmap on the fly when zoomed out, eventually revealing points as you zoom in. But presently this is something you'd have to DIY. For now, clustering is the only thing that works right out of the box.

That’s a useful article, and as impressive as these demos are, and how smooth Maplibre is, I dearly wish people would spend more time on building map tech that is actually based on geoscience.

The heat map is a good example. What does it mean? Points per what? And I can’t bear to talk about the tyranny of Web Mercator.

Maps are so much more powerful when they mean something, and convey scientific facts or actionable evidence.

As far as I know there isn’t a permissively licensed, open source map renderer SDK that can work in a true GIS, e.g. supporting multiple projections.

I still don't understand how the dots help when multiple dots need to be at the exact same coordinates at the lowest zoom level. How do you open a list of them?

I'm working on a project (using Protimaps and MapLibre, even, so this is very timely for me) where users create groups at physical locations. If 12 groups choose the local public library's schedulable meeting room as their meeting space, how would anyone click to see the list of groups that use that one room? Wouldn't I just end up with what looks like one very dark dot?

The clustered markers in leaflet are jarring (I like them, but when I show maps I make my wife she finds the transitions nauseating).

The default heatmaps for these maps are bad. Heatmaps should use filled contours so the gradations are more easily identified. (Continuous raster maps are blobby.) See the ascii glyph map in this post, https://andrewpwheeler.com/2015/06/12/favorite-maps-and-grap.... I think those should be static for various levels of zooms as well, and not recalibrated when zooming.

Another option (not shown here) is to just use polygons and aggregations, and when zoomed in can turn on that point layer (or just have it appear). Or can just make actual clusters (like DBSCAN).

I have a map I made on my website that shows these (with various interaction tooltips/hover), https://crimede-coder.com/graphs/DurhamHotspots (hotspots of crime in Durham, NC). And an explanation of the cartographic decisions and when to use the different techniques, https://www.youtube.com/watch?v=mBm6sTR08BI

Nice article! As someone who implemented both clustering and heatmaps in Mapbox / Maplibre GL JS, just wanted to add that clustering isn't just a visualization technique — it's also a performance feature that can support other visualization types. It might not matter for 30k points, but for e.g. 300k, you can implement dot density with stacked opacity via clustering — use a tiny clustering radius, and style the dots so that opacity is a function of point_count. This creates an approximation of an unclustered viz that performs much smoother. Same thing with heatmaps — you can define heatmap-weight as a function of clustered source's point_count, and it will look nearly identical (aside minor shifts) as the unclustered one, but perform much better.

I really enjoyed this article because it happens to focus on a problem I'm knee-deep in at the moment, although I'm not ready to throw clustering out the window just yet.

There's a deeper issue which the author touches on, showing too much information on a map. This is possible independent of technology limits depending on what level of focus you wish each point to have. For many maps, showing a bunch of small points is an elegant solution, the map turns into more of a data visualization. But a bunch of points tells you very little about a particular thing. It is rich in aggregated information and poor in specific information.

If Google Maps just gave you a bunch of points instead of labels, it would be less functional. Every city would just be a large collection of points. Great, that might be interesting if you want to visualize how the locations are distributed in the region, but you are probably looking at the map because you want to find a particular thing.

Clustering is the natural choice: show the most important thing out of all the points in the area. A label is important, a big circle with a number in it is the worst out all the options discussed. My issue with Google Maps is that there is no visual indicator that there are other points there, and that seems to be largely driven by the fact that it is a marketing platform.

I'm working on a map that uses a hybrid approach. Each map has a narrow focus and is tied to a particular community/interest: concerts, bear sightings, hostels, food trucks, whatever you want. For each point of focus, there is a label and a thumbnail. Showing a bunch of these all in the same place is a bad user experience, so it ranks them and shows the most important one, and the thumbnail is replaced by a number indicating there are other points of focus in the vicinity. Users can click on the item to get a side panel that shows a listing of all the points.

There are also other points on the map that are secondary to its focus, public transportation vehicles and other location points that the user can filter. These are displayed as little dots, similar to the example that the author provides. If you zoom in far enough, they become more than just dots.

In designing the user experience, I tried to make it like a map experience that you might find in a strategy game. Each focus point is like a unit that will bring up more information in a side panel. If there is more than one unit there, show an overview of the units. It's a work in progress, but I'm happy with the result so far.

As somebody who makes maps for a living, I think clusters are overrated (I have had, on multiple occasions, had to talk people out of it). The leaflet extension that gives you a preview of the area covered by the cluster and allows you to click to zoom to the extent of cluster I do feel is a good setup for situations where just putting all the dots down isn't going to give you good information. That being said doesn't work great on mobile.

Also pour one out to google fusion tables which back in the day was the amazing way to get tons of dots onto a map.

I have an app that is heavily map-based. It's an iOS app that uses Apple Maps (I know, I know, controversy, yadda, yadda; but there's a good reason -many of them, in fact- that I use Apple Maps).

I found that the built-in AM clustering truly sucks, so I wrote my own. It's not perfect, but delivers a much better experience than the native one.

>You don’t have to cluster anymore. You can just be.

Try to get Google Maps or Apple Maps running on a phone with more than 200 Markers/Annotations, then come back to me with that. Their performance is fucking dreadful. Google Maps released a new renderer that just OOMs if you're rendering a Polyline with 5k segments. Decimate it or face consequences.

As far as I know, MapBox is about the only one that has tolerable performance. Anyone else doing heavy work and using the gmaps SDK is figuring out tricks: overlay rendering (drawing on a canvas above the map, which requires expensive RPC calls to get visible bounds / map projection which makes performance shit if you're not careful and always lags a frame behind), intense caching of marker descriptors, careful management of markers (dropping 200 Markers from scrolling the map + adding 200? Enjoy your main thread work that freezes the map), etc, etc.

First party map tools are absolute dog shit.

The map in hotels.com does a good job of clustering. For example, searching for hotels in a high density area like NYC. Airbnb, conversely, does not seem to have any clustering. Properties vanish and reappear at various zoom levels.

Flawed article (but not Greg's fault, see Edit).

The dataset he's using is like 10x smaller than the one displayed in AO.

Edit: <@gregsadetsky I have to answer you like this as HN partially censors me by limiting my comments to around 4/day>

Something's weird then (on either end).

Go to Downtown Toronto, I see about 200 places total. On your map I only see about 50.

Actually ... if you scroll down from there, on the southwest side of Lake Ontario, just north of St. Catharines, I see a big cluster of 197 elements, which does not show on your map. I think the map on AO is overcounting markers. You discovered a bug Greg, congrats!

I can see this as a critique on over-use of clustering for whether you have more than a handful of points to visualize. However, I do think this article completely misses the use case when you need to give a ballpark overview of density or counts for clusters. There are many use cases when you actually have natural clusters of points, like traffic accidents. When you zoom out, you might actually see patterns in the aggregated data.

Overall, spatial clustering is a very common strategy to understand trends. To dismiss it entirely is a bit sensational.

This is one of the major gripes that real estate sales agents have.

I’ve consistently heard complaints from customers that they don’t like it. At this point, map technology has come far enough that we could visualize thousands of points using Cesium on consumer-grade hardware so I don’t think it’s a performance issue either.

When it comes to overlapping dots, we generally solve this by displacing the dots minimally so that they’re all still in the same building or same area but visibly different locations.

There are excellent examples of the problem in Edward Tufte's "Envisioning Information". Have a look at http://blah.ksteinfe.com/181106/tufte_envisioning_informatio... for a few snippets of it, then buy the book.

The Atlas Obscura version of clustering seems especially bad. The presence of the cluster circles with unchanged numbers, even after zooming in, is just plain wrong.

I have seen some better clustering implementations, which give good sense for whats inside the cluster if you click on it.

Generally agree that the stacked individual points are a much better approach on modern hardware.

Yeah I hate clustering too. Especially since, like the OP points out, mapping libraries have been easily able to handle thousands of points on screen for many years now.

At least, make the clustering less aggressive so you only get clusters for large numbers of points.

WebGL is a major upgrade for maps. But as an old school web maps person, the obvious solution to the first problem (nearby points clustered even on the city level) is to use the maxZoom option which is in most clustering libraries.

I agree. I hate the clustering when I have to actually work with a map and not just glance at it.

A big offender to me is https://urbantoronto.ca/map - You can zoom in pretty far and still get clusters of 2-5 points that are easily far enough apart that they could be individually seen and clicked.

https://www.realtor.ca/map is another annoying one.

Clustering is only asked for by someone that has no interest in or understanding of the data.